An Oligogenic Case of Severe Neonatal Thrombocytopenia and a Purportedly Benign Variant in GFI1B Requiring Reinterpretation

Abstract

Although thrombocytopenia in neonatal intensive care patients is rarely due to inherited disorders, the number of genetic variants implicated in platelet defects has grown dramatically with increasing genome-wide sequencing. Here we describe a case of severe, oligogenic neonatal thrombocytopenia and reinterpret a reportedly benign mutation that is likely pathogenic. Despite this patient’s synonymous mutation (GFI1B 576 C>T, Phe192=) being annotated as benign, GFI1B is a well-known regulator of megakaryopoiesis, this variant alters splicing and megakaryocyte maturation, and our analysis of existing genome-wide associated studies demonstrate that it likely causes gray platelet syndrome. This variant has not been reported in a case of life-threatening thrombocytopenia. We propose the severity of this patient’s phenotype is due to synergistic epistasis between the intrinsic platelet defect caused by this mutation and her concomitant inherited PMM2 congenital glycosylation disorder neither of which have been associated with such a severe phenotype. This case highlights the importance of whole exome/genome sequencing for critically ill patients, re-examining variant interpretation when clinically indicated, and the need to study diverse genetic variation in hematopoiesis.

BACKGROUND

Inherited disorders are a seemingly infrequent cause of severe thrombocytopenia in neonatal intensive care patients. However, genetic diseases are likely an underappreciated subset of platelet disorders given that 50–80% of variance in platelet counts is heritable^1,2 and that the cause of clinically significant thrombocytopenia is often not identified. We present an oligogenic case of severe thrombocytopenia in a neonate and reinterpret a reportedly benign synonymous substitution in GFI1B (576 C>T, Phe192=) as a causal variant. We argue that this patient’s severe phenotype is a result of synergistic epistasis between the intrinsic platelet defect caused by GFI1B 576 C>T and consumptive coagulopathy caused by her inherited PMM2 congenital disorder of glycosylation (PMM2-CDG). Neither mutation has previously been associated with life threatening thrombocytopenia. This case illustrates the importance of unbiased deep sequencing in neonatal thrombocytopenia, the genetic heterogeneity of platelet disorders, the need to reevaluate known variants given new data, and the profound effect of epistasis among multiple causal variants.

CASE PRESENTATION

A baby girl was born prematurely at 28w6d to a 39 year-old G2P0 mother by Cesarean section because of difficulty with extraction in the setting of preeclampsia with severe features. The pregnancy was complicated by gestational diabetes mellitus and maternal Hashimoto’s thyroiditis that was well-controlled with levothyroxine. The patient was diagnosed with hydrops fetalis on a 26-week fetal ultrasound and prenatal karyotyping was normal. After delivery, the patient was admitted to the neonatal intensive care unit (NICU) for management of severe thrombocytopenia, hydrops fetalis, and prematurity. Petechiae and bruising were noted at birth, and blood was seen during intubation. Her initial platelet count was ~6,000 /uL (reference 160,000 – 370,000 /uL). While platelets were sparse, most of those that were seen on peripheral smear were large, hypogranular, and were not aggregated (Figure 1). She was transfused, responded robustly (>100,000 platelets/uL), and was treated empirically with intravenous immunoglobulin (IVIG). However, all serologies returned negative, and her thrombocytopenia persisted over 72 hours.

Figure 1:

Peripheral blood smear from the patient at the time of birth demonstrating sparse and hypogranulated macrothrombocytes (black arrows). Insert: magnified hypogranular macrothrombocytes from bottom left of widefield microscopy. RBC volume (from CBC) and morphology are within normal limits. Her RBC diameter therefore is also normal (~8 μm). Platelets shown are often 50% or more of RBC diameter making the platelet diameters ~4 μm or larger and therefore above the normal 1.5–3 μm range.

Common causes of persistent neonatal thrombocytopenia were ruled out with negative maternal infectious studies, blood cultures, and X-rays (excluding thrombocytopenia absent radius syndrome). Thrombopoietin (TPO) was severely elevated at 3,692 pg/mL (reference interval 7 – 99 pg/mL), but a bone marrow biopsy could not be performed given that the patient weighed ~2.5 kg. She was transfused dozens of times during the 5-month hospital course, which maintained an average platelet count of ~45,000 /uL (range 15,000 – 156,000 /uL). All other blood cell counts (including red blood cell counts) were unremarkable and there were no episodes of clinically significant bleeding. Her PTT and INR were mildly elevated at averages of 53 (range 43.1 – 67.1) and 1.4 (range 1.1 – 1.8), respectively.

In addition to her severe thrombocytopenia, she had anasarca, required therapeutic paracentesis, bilateral chest tubes, and repeated albumin and furosemide infusions. Given these multisystem defects and thrombocytopenia of unknown etiology, whole exome sequencing was performed. This revealed compound heterozygosity for two known pathogenic mutations in PMM2 (Phe119Leu and Phe157Ser) and heterozygosity for a synonymous substitution in GFI1B (576 C>T, Phe192=, ClinVar ID 711822), which was found to be inherited from her mother. The mother’s CBC was normal with a platelet count of 227,000 /uL and mean platelet volume of 12.4 fL. The patient was diagnosed with PMM2-congenital disorder of glycosylation (PMM2-CDG), and she received several fresh frozen plasma infusions, bumetanide, and a two-day course of protein C concentrate for diffuse endothelial barrier dysfunction³. Her native protein C activity level was depressed at <10% normal pooled plasma (NPP). The normal adult range is 70–150% NPP. However, at birth, protein C activity is (on average) approximately 35% adult value making it difficult to interpret this patient’s low activity level. After receiving two days of protein C concentrate, repeat measurement showed an increase to 45.2% NPP. Her hospital course was also complicated by an occlusive thrombus in her right lower leg at the site of insertion for a central catheter that was successfully treated with a 2-week heparin drip and resolved on repeat ultrasound. Unfortunately, she became septic in the 5^th month of life, rapidly decompensated, interventions were withdrawn, and she died.

DISCUSSION

We present a case of severe, oligogenic neonatal thrombocytopenia and reinterpret a benign variant as pathogenic. PMM2-congenital disorder of glycosylation (CDG) is unlikely to be the sole cause of this patient’s life-threatening thrombocytopenia given that PMM2-CDG is rarely associated with thrombocytopenia⁴. Moreover, there are no reports of patients with PMM2-CDG requiring near daily platelet transfusions to survive. While case reports exist of PMM2-congenital disorder of glycosylation (PMM2-CDG) with thrombocytopenia^5,6, the predominant defect in this disorder is that of secondary hemostasis. Further, these cases exist in the rare subset of patients with PMM2-CDG who survive into adulthood⁶ (a phenotypically and likely biomolecularly distinct entity from the typical, severe PMM2-CDG), these cases did not use unbiased whole genome sequencing to exclude contributing genetic confounders, and pale macrothrombocytes have never been associated with the disorder. Theoretically, PMM2-CDG could cause a primary hemostatic defect either by platelet consumption secondary to endothelial barrier dysfunction or by impaired platelet surface protein glycosylation. Neither can account for the entirety of this patient’s extreme phenotype. Endothelial barrier dysfunction and hydrops is common in PMM2-CDG whereas thrombocytopenia is not arguing against PMM2-CDG-mediated barrier dysfunction and platelet consumption as the sole cause of her platelet deficiency. The latter is possible given that platelets from patients with PMM2-CDG have reduced sialic acid content though there is no clear quantitative or qualitative glycosylation defect in platelets from patients with PMM2-CDG⁷. Taken together, PMM2-CDG alone is unlikely to account for the entirety of this patient’s severe macrothrombocytopenia.

On the other hand, the mutation GFI1B 576 C>T is a prime candidate for being causal despite being annotated as benign (ClinVar ID 711822). GFI1B is a transcriptional repressor critical for megakaryopoesis^8,9 and several pathogenic GFI1B variants have been associated with impaired platelet production and function^10–12. GFI1B 576 C>T is a synonymous substitution (Phe192=) that alters splicing to produce a short isoform¹³ that inhibits megakaryocyte differentiation (while sparring erythropoiesis) in a dominant-negative fashion¹⁴ which explains this patient’s phenotype despite being heterozygous. A megakaryocyte differentiate block also explains this patient’s remarkably elevated thrombopoietin (3,692 pg/mL, reference interval 7 – 99 pg/mL). While we couldn’t confirm a defect in her megakaryocyte lineage without a bone marrow aspirate, this patient’s exceedingly high thrombopoietin (TPO) suggests defective differentiation is possible. Although elevated TPO could indicate increased platelet turnover in the setting of impaired platelet glycosylation, platelet glycosylation defects have never been definitively confirmed⁷. Additionally, such severely elevated TPO (>30 times the normal upper limit) has never been reported in CDG but is commonly observed in congenital amegakaryocytic thrombocytopenia (CAMT), a disorder of impaired megakaryopoesis. In CAMT, TPO is severely elevated given the lack of TPO-receptor (c-Mpl) expressing megakaryocytes which would otherwise remove TPO from circulation. It is likely that this patient’s elevated TPO is due to combination of impaired megakaryocyte differentiation and possibly impaired c-Mpl glycosylation which could inhibit signaling through c-Mpl.

There is also significant population-scale data to implicate GFI1B 576 C>T in this patient’s thrombocytopenia. Several genome-wide association studies (GWAS) in adult populations have found that this variant is highly associated with significant increased platelet volume and decreased platelet count but not associated with erythrocyte count^15–18 (Table 1). Indeed, this patient’s sparse thrombocytes are large (Figure 1). The existing functional and population-level sequencing data are consistent with this patient’s macrothrombocytopenia and normal RBC counts, and suggest that GFI1B 576 C>T be reclassified as pathogenic based on American College of Medical Genetics criteria (PVS1, PS3, PS4, PM1, PM4)¹⁹.

Table 1:

GFI1B 576 C>T is highly associated with quantitative platelet traits in diverse adult populations. Data listed is derived from previously published genome-wide association studies.

Trait	Direction	p-value	Beta (SD)	CI	Study
Platelet Count	Decrease	2 × 10−292	0.41	0.39 – 0.43	Chen et al. Cell. 2020
Platelet Count	Decrease	4 × 10−232	0.43	0.41 – 0.46	Sakaue et al. Nat Gen. 2021
Platelet Count	Decrease	5 × 10−198	0.41	0.38 – 0.43	Vuckovic et al. Cell. 2020
Platelet Count	Decrease	6 × 10−88	0.42	0.38 – 0.46	Astle et al. Cell. 2016
Platelet Count	Decrease	1.8 × 10−27	0.43	0.34 – 0.52	Polfus et al. AJHG. 2016
Platelet Volume	Increase	1 × 10−255	0.46	0.44 – 0.49	Vuckovic et al. Cell. 2020
Platelet Volume	Increase	2 × 10−107	0.47	0.43 – 0.51	Astle et al. Cell. 2016
Platelet Distribution Width	Increase	3 × 10−74	0.25	0.22 – 0.27	Vuckovic et al. Cell. 2020
Platelet Distribution Width	Increase	4 × 10−40	0.28	0.24 – 0.32	Astle et al. Cell. 2016

In addition to her quantitative platelet defect, this patient’s large and hypogranulated platelets are qualitatively reminiscent of gray platelet syndrome (GPS). Intriguingly, an autosomal-dominant GPS has been described where the causal defect is a heterozygous nonsense mutation in GFI1B¹¹. That GFI1B 576 C>T is associated with macrothrombocytopenia (Table 1) and that this patient had hypogranulated macrothrombocytes leads to the suggestion that GFI1B 576 C>T might cause a disease that exists on a continuum with gray platelet syndrome. Alternatively, the pale granulations seen on this patient’s peripheral smear may indicate that platelets have been activated. While this is likely contributing, we hypothesize that impaired megakaryocyte differentiation caused by GFI1B 576 C>T may result in a similar phenotype given the existing in vitro evidence that GFI1B 576 C>T inhibits megakaryocyte differentiation, that additional dominant-negative GFI1B variants alter megakaryocyte morphology¹¹, and given this patient’s corroborating severely elevated TPO.

This is the first reported case of GFI1B 576 C>T in a neonate and the first report of the variant as pathogenic. This is also the most severe phenotype reported with this variant; notably, the patient’s mother has the same GFI1B genotype but a normal platelet count, normal MPV, and no clinically apparent bleeding disorder. A mild or non-apparent hematologic defect with GFI1B 576 C>T alone is further supported by the many instances of this genotype in large cohorts of adults (Table 1). This patient’s life-threatening thrombocytopenia is therefore likely of oligogenic origin: epistasis between the patient’s intrinsic platelet defect driven by GFI1B 576 C>T and her concomitant glycosylation defect. We illustrate the importance of unbiased whole exome/genome sequencing in neonatal thrombocytopenia, the genetic heterogeneity of platelet disorders, the role of oligogenic causes in rare diseases, and the need to reevaluate known variants given new data. GFI1B alterations should be routinely included in the differential diagnosis for macrothrombocytopenia.

PLAIN LANGUAGE SUMMARY

What is the context?

Low platelets (thrombocytopenia) in the neonatal population isn’t frequently inherited. As we perform unbiased DNA sequencing in more patients, the number of inherited platelet disorders and implicated variants are growing.

The gene GFI1B encodes for a transcription factor that regulates megakaryocytes, the cell type that produces platelets. A synonymous substitution in GFI1B (576 C>T, Phe192=) is annotated as benign; however, experimental studies have shown that it inhibits megakaryocyte production.

There is growing appreciation for oligogenic inheritance where multiple causal variants contribute to clinical phenotypes.

What is new?

We present a case of life-threatening neonatal macrothrombocytopenia (large, hypogranulated sparse platelets) that has an oligogenic cause. We reinterpret the synonymous substitution GFI1B 576 C>T as pathogenic.

This patient’s severe phenotype was likely due to the combined effect of GFI1B 576 C>T and her inherited glycosylation disorder (PMM2-CDG). Neither variant alone causes severe thrombocytopenia, but the combined intrinsic platelet defect (GFI1B mutation) and consumption (PMM2-CDG) likely produced her life-threatening phenotype.

What is the impact?

GFI1B is a critical regulator of megakaryocyte production. The purportedly benign mutation 576 C>T is likely pathogenic causing thrombocytopenia by impairing megakaryocyte maturation.

As more patients have unbiased genome sequencing, oligogenic and polygenic inheritance will become increasingly appreciated as causes of platelet disorders.

NICU providers should consider whole genome or exome sequencing of neonates with severe thrombocytopenia after reversible causes are ruled out.

LIST OF ABBREVIATIONS

PMM2

Phosphomannomutase 2

PMM2-CDG

PMM2 – Phosphomannomutase 2 congenital disorder of glycosylation

pLoF

Predicted loss of function

GPS

Gray platelet syndrome

GFI1B

Growth factor independent 1B transcriptional repressor

NBEAL2

Neurobeachin-like 2

TPO

Thrombopoietin

PTT

Partial thromboplastin time

INR

International normalized ratio

AST

Aspartate aminotransferase

ALT

Alanine transaminase

IVIG

Intravenous immunoglobulin

NICU

Neonatal intensive care unit

AVAILABILITY OF DATA AND MATERIALS

Data sharing is not applicable to this article as no datasets were generated during the current study.